Visual Based Study On Eye Movement

Eyes are the most important sensory organs of mankind,and the most striking feature of human face.The research of eye movement plays a key role in understanding human visual work mechanisms,understanding human emotions and behaviors,and eye-based human-computer interaction,fatigue driving,and so on.The human eye detection and tracking technology based on computer vision have the advantages of low intrusion and high precision,and it has become the mainstream direction of eye movement research.In the past few decades,although the eye movement characteristics of research has made significant progress,but there are still many uncontrollable factors,affecting the test results.The shape and size of each person's eye is different,and the area of the eyelids to block the eye is not the same,sometimes location and lighting conditions is different,also there are eyelashes and other noise interference which will directly affect the accuracy of eye movement research.Now,the price of equipment that can collected eye data is very expensive and scarce in the market.So most of the research of eye movement characteristics are based on the face detection,and the direct detection of the eye and the feature analysis are very scarce.In view of these problems,this paper focuses on the eyelid and pupil eye movement characteristics.In this paper,a new eye-detection AK-EYE model based on ASM algorithm and Kalman filter is proposed,and the model is used to improve the speed and precision of eyelid matching,and to improve the tracking algorithm about the shape and position to the eye.Then,we use matching technique to locate the pupil,and collect the eye data set to verify the experiment.Finally,we analyze the fatigue detection method with eye movement characteristics.The main contents are as follows:(1)In this paper,the main ideas,advantages and disadvantages of the related eye movement feature research methods are analyzed in detail.In the process of eyelid matching and positioning,aiming at the problem of misalignment in ASM algorithm,this paper proposes a AK-EYE model which combines ASM algorithm and Kalman filter,and use the model to match the eyelids.First,we select the appropriate sample characteristics according to the specific criteria for calibration and establish eyelid shape model.Then,we use the eyelid shape model to match the eyelid contour,also to predict and update the initial location of the model to complete the real-time eyelid positioning tracking,and analyze the experimental results.(2)This paper chooses the appropriate template,and uses the template matching algorithm to achieve the precise positioning of the pupil for the data set of the human face and the data set of the simple eye,and then analyzes the experimental results.Finally,we compare the performance of different template matching algorithms,and apply the template matching algorithm to public database for effect verification.(3)In this paper,we study the eyelid information and pupil information that obtained from the eye movement characteristics to establish the fatigue driving warning system.The fatigue warning system uses the eyelid feature information and the pupil characteristic information as the input parameters of the fatigue state,and combines the eye movement characteristics and the PERCLOS determination principle to construct a new fatigue analysis and judgment algorithm to judge the fatigue state.